沉默的科学家：当软件研究未能触及其受众

好的，这是根据您提供的内容整理的Markdown格式中英双语文章。

# The Silent Scientist: When Software Research Fails to Reach Its Audience
# 沉默的科学家：当软件研究未能触及其受众

### Research impact starts with science communication.
### 研究的影响力始于科学传播。

---

For researchers, reflecting on their own work is part of good scientific practice. This is common across all scientific disciplines, including software research.a Though critical self-reflection is a staple in research, the software research community seems to take it further, often grappling with feelings of insignificance and questioning the relevance of their work. In the past two years, such discussions have been nearly impossible to track. Two prominent examples illustrate this: In 2022, Lionel Briand delivered a keynote at ICSE, the largest software engineering conference, critically discussing the impact we can expect from software research.b Two years later, one of the leading software journals, the *Journal of Systems and Software,* launched a new column inviting software practitioners to share their perspective on making software research more relevant.1 Apparently, the software research community has reached a new peak in its ongoing quest to doubt its own relevance and impact.

对于研究人员来说，反思自己的工作是良好科学实践的一部分。这在包括软件研究在内的所有科学学科中都很常见。a 尽管批判性的自我反思是研究中的主要内容，但软件研究社区似乎更进一步，常常在无足轻重的感觉中挣扎，并质疑其工作的相关性。在过去两年里，这样的讨论几乎无法追踪。两个突出的例子说明了这一点：2022年，Lionel Briand在最大的软件工程会议ICSE上发表了主旨演讲，批判性地讨论了我们可以从软件研究中期待的影响。b 两年后，顶尖的软件期刊之一《系统与软件杂志》（*Journal of Systems and Software*）推出了一个新专栏，邀请软件从业者分享他们关于如何使软件研究更具相关性的观点。1 显然，软件研究界在不断质疑自身相关性和影响力的过程中达到了一个新的高峰。

To some extent, we understand the sentiment. Significant resources go into software research, to improve collaboration or drive innovation. It is reasonable to critically question how many of these research findings will be implemented in the near or distant future. In fact, considerable resources have already been devoted to addressing this very question, and analyses have identified concrete examples showing how software research contributed to advancements in development tools and methods, such as configuration management and programming languages.c

在某种程度上，我们理解这种情绪。大量的资源投入到软件研究中，以改善协作或推动创新。批判性地质疑这些研究成果中有多少将在近期或遥远的未来得到实施是合理的。事实上，已经有相当多的资源用于解决这个问题，并且有分析已经确定了具体的例子，说明软件研究如何促进了开发工具和方法的进步，例如配置管理和编程语言。c

However, in the ongoing discussions surrounding the relevance and impact of software research, there is a notable oversight: the crucial role of science communication. It is surprising how often this aspect is disregarded. Like a silent scientist, quietly publishing papers and expecting their work to speak for itself, many researchers assume their findings will naturally find their audience. This assumption can lead to an overly self-critical misconception: that if research lacks impact, it must be irrelevant. In this essay, we set out to challenge this view. Research can only have impact if it reaches its target audience in the first place, which requires making findings accessible through various communication channels.8,15 The tricky thing about this situation is that, so far, active science communication has been sparse in the area of software research, and those who have tried often find their efforts unrewarded or unsuccessful.13

然而，在围绕软件研究的相关性和影响力的持续讨论中，有一个明显的疏忽：科学传播的关键作用。令人惊讶的是，这个方面经常被忽视。就像一个沉默的科学家，悄悄地发表论文并期望其工作能不言自明，许多研究人员认为他们的发现会自然而然地找到受众。这种假设可能导致一种过度的自我批判误解：如果研究缺乏影响力，那它一定是无关紧要的。在本文中，我们旨在挑战这一观点。研究只有在首先触及其目标受众时才能产生影响，这需要通过各种传播渠道使研究成果易于获取。8,15 这种情况的棘手之处在于，到目前为止，在软件研究领域，积极的科学传播还很稀少，而那些尝试过的人往往发现他们的努力没有得到回报或未获成功。13

## Understanding the Relevance of Software Research
## 理解软件研究的相关性

To appreciate the relevance of software research, we need to consider how different topics resonate with different stakeholder groups—what is impactful to one may be irrelevant to another.

要理解软件研究的相关性，我们需要考虑不同主题如何与不同的利益相关者群体产生共鸣——对一方有影响的内容可能对另一方毫无意义。

First, software research covers diverse content, which can be roughly divided into technical advancement and (human-centered) empirical understanding. It seeks technical improvements, such as methods to automatically find and fix bugs, and empirical insights, such as understanding what factors influence the productivity of software developers. Corresponding studies of the two types of research use entirely different research methodologies, with some focusing on technical evaluations without involving humans, while others include human participants as primary subjects. This alone can create varied perceptions of relevance, as some studies directly involve the people they aim to help. Additionally, measuring impact and implementing findings differ significantly. For technological progress, impact is measured through adoption or metrics on quality and performance. Findings on practitioner collaboration require integration into sociotechnical processes, with impact measured through changes in behavior or improved satisfaction.

首先，软件研究涵盖了多样化的内容，大致可以分为技术进步和（以人为中心的）经验性理解。它寻求技术改进，例如自动发现和修复错误的方法；也寻求经验性见解，例如理解哪些因素影响软件开发人员的生产力。这两种类型的研究使用完全不同的研究方法，一些侧重于不涉及人类的技术评估，而另一些则将人类参与者作为主要研究对象。仅此一点就可能产生对相关性的不同看法，因为一些研究直接涉及它们旨在帮助的人群。此外，衡量影响和实施研究成果的方式也大不相同。对于技术进步，影响力通过采纳率或质量和性能指标来衡量。而关于从业者协作的研究成果则需要融入社会技术流程，其影响力通过行为改变或满意度提高来衡量。

Second, different phases of the research process are of interest to different stakeholders. A research project can span several years and progress through various phases, at the end of which one or more publications may appear. Judging the relevance of research based on a single publication inevitably leads to a large proportion of the readership not (yet) feeling addressed. Let us assume we are looking for a source code quality metric that indicates the comprehensibility of source code. Thanks to software research, we know that most code comprehensibility metrics do not, in practice, reflect what they are supposed to measure.9,10 So, we are well advised not to repeat the mistake and simply design another metric based on our gut feeling. We start a little earlier, define the notion of code comprehension, conduct basic research, and understand the neuropsychological correlates of code comprehension in the brain of developers to develop a meaningful metric in a subsequent step.11 Will that conceptual definition of code comprehension or neurophysiological lab studies with small code snippets hold actionable insights for practice? Probably not. Are these studies relevant at all? Absolutely, because software research has more than just the target group of software practitioners. In this case, these early findings may help educators better support novice programmers with code comprehension difficulties, guide experimenters to design better studies and ask more targeted questions, and they eventually aid other researchers build their application-oriented research on a solid theoretical foundation.

其次，研究过程的不同阶段对不同的利益相关者具有不同的吸引力。一个研究项目可能持续数年并经历多个阶段，最终可能产生一篇或多篇出版物。仅根据单篇出版物来判断研究的相关性，必然会导致大部分读者感觉自己（还）不是目标受众。假设我们正在寻找一个能够指示源代码可理解性的质量度量标准。得益于软件研究，我们知道大多数代码可理解性度量标准在实践中并不能反映其应有的衡量效果。9,10 因此，明智的做法是不要重蹈覆辙，仅仅凭直觉设计另一个度量标准。我们从更早的阶段开始，定义代码理解的概念，进行基础研究，并了解开发者大脑中代码理解的神经心理学关联，以便在后续步骤中开发出有意义的度量标准。11 那么，这种代码理解的概念定义或使用小代码片段的神经生理学实验室研究，对实践是否具有可操作的见解呢？可能没有。这些研究到底有没有相关性呢？绝对有，因为软件研究的目标群体不仅仅是软件从业者。在这种情况下，这些早期发现可能帮助教育工作者更好地支持在代码理解上有困难的新手程序员，指导实验者设计更好的研究并提出更有针对性的问题，并最终帮助其他研究人员在坚实的理论基础上建立其面向应用的研究。

The point is that relevance is much more complex than is often portrayed. What is irrelevant to one person today may be very relevant for another tomorrow. We assure the reader that this is not just a convenient excuse that allows researchers to retreat into an ivory tower. We ourselves see systemic difficulties that reinforce the impression of a lack of relevance of software research. For example, the very people we want to help with our research at any given time are often involved too late in the research process. Researchers who have not spoken and, in particular, listened to their target group risk creating an artificially constructed problem space. Listening can take many forms, such as attending industry meetups, conducting targeted surveys, or monitoring developer forums and social media. Without these insights, it becomes difficult to argue for the relevance of one’s research, and even harder to attract anyone as an audience for science communication. Another challenge is certainly that software researchers are incentivized by the scientific review and funding process to present solutions to be as generalizable as possible. As a result, we hear of software practitioners who fail to adapt published research findings because research has raised false expectations of applicability.

关键在于，相关性远比通常描述的要复杂得多。今天对一个人无关紧要的事情，明天可能对另一个人非常重要。我们向读者保证，这并非只是让研究人员退回象牙塔的方便借口。我们自己也看到了系统性的困难，这些困难加深了软件研究缺乏相关性的印象。例如，我们在研究中想要帮助的那些人，往往在研究过程中被卷入得太晚。没有与目标群体交谈，特别是倾听他们意见的研究人员，可能会创造一个被人为构建出来的问题空间。倾听可以有多种形式，如参加行业聚会、进行有针对性的调查，或监控开发者论坛和社交媒体。没有这些见解，就很难论证自己研究的相关性，更难吸引任何观众进行科学传播。另一个挑战当然是，软件研究人员在科学评审和资助过程中，被激励去提出尽可能具有普适性的解决方案。结果，我们听说有软件从业者因为研究提出了不切实际的适用性期望，而无法调整和应用已发表的研究成果。

Note that, taking this viewpoint, we will neither succeed in fundamentally rethinking the academic system nor in convincing every reader of the relevance of every single research paper. We do not even want to. There are already enough opinions on why some software research is theoretical in nature and why other research should be more application-oriented. There is a multitude of opinions on what role industry–academia collaborations could play for the relevance of research topics. And there are certainly enough opinions on how the impact of software research could be evaluated. Our point is different: As long as the research community does not manage to make its research accessible to the respective target group at all, we do not even need to talk about relevance and impact.

请注意，持此观点，我们既无法成功地从根本上反思学术体系，也无法说服每一位读者相信每一篇研究论文的相关性。我们甚至不想这样做。关于为什么一些软件研究本质上是理论性的，而另一些研究应该更面向应用，已经有足够多的观点。关于产学研合作在研究课题相关性方面可以扮演什么角色，也有大量的看法。当然，关于如何评估软件研究的影响，也有足够多的意见。我们的观点不同：只要研究界无法使其研究成果为相应的目标群体所接触，我们甚至无需讨论相关性和影响力。

## No Impact without Science Communication
## 没有科学传播就没有影响力

The feeling of irrelevance and missing impact can often be traced back to a communication problem: The reality is that nobody cares about your research unless you make them care. This does not happen by itself with the publication of a research paper. The notion that software practitioners will dive into the academic world, eagerly browse online libraries for papers, and read them with sparkling eyes is as romantic as it is unlikely. A far more probable scenario for success is when researchers and practitioners build a bridge between these two worlds through actively engaging in dialogue.

无关紧要和缺乏影响力的感觉，往往可以追溯到一个沟通问题：现实是，除非你让他们关心，否则没人会在意你的研究。这并不会随着一篇研究论文的发表而自然发生。认为软件从业者会一头扎进学术界，热切地在在线图书馆中浏览论文，并以闪亮的眼睛阅读它们，这种想法既浪漫又不切实际。一个更有可能成功的场景是，研究人员和从业者通过积极参与对话，在这两个世界之间架起一座桥梁。

The fact that software practitioners are indeed interested in contemporary software research is shown by a recent study that examines how these research findings are disseminated and discussed on LinkedIn.14 This data reveals the majority of individuals who post content and comment on the posts are software practitioners. The authors of that study conclude that researchers are not doing enough themselves. They note that some software research is so engaging that practitioners take on the role of science communicators—a role traditionally reserved for researchers, but inadequately filled by them in the case of software research.14

最近一项研究考察了当代软件研究成果如何在LinkedIn上传播和讨论，结果表明软件从业者确实对此感兴趣。14 数据显示，发布内容和评论帖子的大多数人都是软件从业者。该研究的作者得出结论，研究人员自身做得还不够。他们指出，一些软件研究非常有吸引力，以至于从业者承担起了科学传播者的角色——这个角色传统上是为研究人员保留的，但在软件研究领域，他们并未充分履行这一职责。14

The good news is that researchers can do a lot to improve science communication without having to revamp the publication system or redesign research processes. If we follow Burns et al.4 in their definition of science communication “as the use of appropriate skills, media, activities, and dialogue to produce one or more of the following personal responses to science: Awareness, Enjoyment, Interest, Opinion-forming, and Understanding,” it is difficult to find a software research paper that cannot be brought to the target group with, at least, one of these intentions. Therefore, post-publication science communication is something every researcher can and should engage in.

好消息是，研究人员可以在不彻底改革出版体系或重新设计研究流程的情况下，为改善科学传播做出很多努力。如果我们遵循 Burns 等人4 对科学传播的定义，即“使用适当的技能、媒体、活动和对话，以产生以下一种或多种对科学的个人反应：认知、享受、兴趣、观点形成和理解”，那么很难找到一篇软件研究论文不能以至少其中一种意图传达给目标群体。因此，发表后的科学传播是每位研究人员可以也应该参与的事情。

Why is it not happening? We see two main reasons. First, publications are the currency of academia. Career advancement largely depends on the number of top-tier publications, while outreach and the practical impact of research are often secondary in university selection processes. Considering such systemic incentives, it is understandable that researchers quickly move on to the next project after publishing.

为什么这种情况没有发生？我们看到两个主要原因。首先，出版物是学术界的硬通货。职业发展在很大程度上取决于顶级出版物的数量，而外展活动和研究的实际影响在大学的选拔过程中往往是次要的。考虑到这种系统性激励，研究人员在发表论文后迅速转向下一个项目是可以理解的。

Second, there is a lack of evidence on the effectiveness of science communication in software research, so it may be useful to look at other sciences. Bauer et al.’s introduction to implementation science, that is, “a science of implementation,” is only 10 years old, but has since been cited more than 2,000 times, many times by successful field reports.2 In the context of clinical research, “the relatively new field of implementation science has developed to enhance the uptake of evidence-based practices and thereby increase their public health impact.”3 Software research needs similar initiatives to learn how to bring research findings into software practice. In contrast, empirical insights on how research transfer works in software engineering and how science communication actually affects software engineering practice are still missing. As long as this gap persists, it is challenging to convince researchers that the extra effort is worthwhile.

其次，关于科学传播在软件研究中有效性的证据不足，因此借鉴其他科学领域可能有所帮助。Bauer等人对实施科学（即“一门关于实施的科学”）的介绍虽然只有10年历史，但此后已被引用超过2000次，其中许多来自成功的实地报告。2 在临床研究的背景下，“相对较新的实施科学领域已经发展起来，以促进循证实践的采纳，从而增加其公共卫生影响。”3 软件研究需要类似的举措来学习如何将研究成果引入软件实践。相比之下，关于研究转化在软件工程中如何运作以及科学传播如何实际影响软件工程实践的经验性见解仍然缺失。只要这一差距存在，就很难说服研究人员付出额外的努力是值得的。

In any case, what is difficult to dispute is that research is unlikely to have much impact without any communication. This can take the form of a blog post, a social media discussion, a workshop or Dagstuhl seminar involving the target audience, or any other channel that fits the audience.5,6 ACM itself offers several channels for this very purpose, such as the *Communications* BLOG@CACM column and *ACM Queue*. Even something as simple as submitting a ticket in an open source project that you have studied in your research can help connect your work to those who would benefit. While the software research community may not yet know whether these efforts will lead to new collaborations, more citations, or industry-wide change, there is reason to assume that effective communication can make a difference. To provide, at least, anecdotal encouragement, we quote software researcher Marcos Kalinowski, who wrote on social media:7 “Recently I shared the result of a Ph.D. thesis on LinkedIn and it reached 4,000+ reactions and 270,000+ impressions. 95% of my network is from industry. We are well equipped to burst the academic bubble!”

无论如何，难以否认的是，没有任何沟通的研究不太可能产生太大影响。沟通可以采取多种形式，如博客文章、社交媒体讨论、有目标受众参与的研讨会或达格施图尔（Dagstuhl）研讨会，或任何其他适合受众的渠道。5,6 ACM本身也为此提供了多个渠道，例如《*Communications*》的BLOG@CACM专栏和《*ACM Queue*》。即使是像在你研究过的开源项目中提交一个工单（ticket）这样简单的事情，也可以帮助你将工作与那些可能受益的人联系起来。虽然软件研究界可能还不知道这些努力是否会带来新的合作、更多的引用或行业范围的变革，但有理由相信，有效的沟通可以带来改变。为了提供至少是轶事性的鼓励，我们引用软件研究员Marcos Kalinowski在社交媒体上写道：7 “最近我在LinkedIn上分享了一篇博士论文的成果，获得了4000多个反应和超过27万次展示。我95%的人脉来自业界。我们完全有能力打破学术泡沫！”

Looking ahead, researchers are well advised to ensure their work reaches the right audience by complementing traditional, high-quality academic paper publishing with efforts to make research more accessible to relevant stakeholders. Identify your target audience to tailor your message! Use diverse communication channels beyond papers, and actively engage with practitioners to foster dialogue rather than broadcasting information! This approach can spark feedback, opening doors to new ideas and collaborations that shape future research. If you are in software research, or any other research area for that matter, remember why you began your research journey. The silent scientist may publish papers, but without reaching the right audience, this work risks going unnoticed and unappreciated. We therefore urge researchers to break the silence and actively approach those communities that stand to benefit most from the findings.

展望未来，我们强烈建议研究人员在发表传统、高质量的学术论文之外，努力使研究成果更容易为相关利益相关者所接触，以确保他们的工作能够触及正确的受众。识别你的目标受众以定制你的信息！使用超越论文的多样化沟通渠道，并积极与从业者互动以促进对话，而非单向传播信息！这种方法可以激发反馈，为塑造未来研究的新想法和合作打开大门。如果你从事软件研究或任何其他研究领域，请记住你开始研究之旅的初衷。沉默的科学家或许能发表论文，但如果无法触及正确的受众，这项工作就有可能被忽视和不被赏识。因此，我们敦促研究人员打破沉默，积极接触那些最能从研究成果中受益的社群。